AnsweredAssumed Answered

How to optimize REST API usage?

Question asked by jy0043976 on Mar 5, 2018
Latest reply on Mar 6, 2018 by jy0043976

Like • Show 0 Likes0
Comment • 2

Our organization has written a number of B2s which integrate with Blackboard via the Java APIs. Now we're working on a rewrite of an old product, switching from a B2 to an LTI, using the REST APIs. Most of our products are tools for teachers, so frequently we'll pull all the Course Memberships for a course, then all the users, etc. From what I can tell, the Bb REST API doesn't support batching (asking for multiple users at the same time) or embedding (where a user object appears inside the Course Membership object).

We're making a large number of REST API calls to load one course, especially if that course has 500+ Course Memberships:

We make a call for the Course Memberships
For each Course Membership we make a call for the user
Then we'll make other calls for user specific data (grades, etc), one (or more) calls for each Course Membership.

Beyond batching or embedding, are there other ways that we can improve performance?

Thanks!

No one else has this question

Outcomes

scott.hurrey Mar 5, 2018 4:27 PM

Your analysis is correct, Jeff. My suggestion would be to cache or store the user data, so when you get memberships, you can just pull in new users and drop users that are no longer enrolled. Still, with your enrollment numbers, there will be a large number of calls, but caching should help a lot.

1 person found this helpful

Actions

jy0043976 @ scott.hurrey on Mar 6, 2018 6:56 AM

Thanks, Scott! Caching is something we're definitely planning to use. I was hoping that I was missing something and that there were other ways to optimize performance. :-)

From what I can see, the Bb REST API is correctly adhering to REST standards, but I have used other APIs (Canvas') that extended REST a bit, or gave alternative ways to get data in larger batches. For example, something like /learn/api/public/v1/courses/{courseId}/users that would return user information for all users in a course in one REST call. Or possibly a way to embed User information inside each Course Membership (this does violate REST, though). With Canvas, we found that by leveraging these other alternative methods we improved performance considerably. With the {courseId}/users example, we could make simultaneous requests for Course Memberships and Users, then merge them, instead of waiting for the Course Memberships call to return to request Users.

Actions

2 Replies

scott.hurrey Mar 5, 2018 4:27 PM
Mark Correct
Correct Answer
Your analysis is correct, Jeff. My suggestion would be to cache or store the user data, so when you get memberships, you can just pull in new users and drop users that are no longer enrolled. Still, with your enrollment numbers, there will be a large number of calls, but caching should help a lot.
1 person found this helpful
Like • Show 0 Likes0
Actions
- jy0043976 @ scott.hurrey on Mar 6, 2018 6:56 AM
  Mark Correct
  Correct Answer
  Thanks, Scott! Caching is something we're definitely planning to use. I was hoping that I was missing something and that there were other ways to optimize performance. :-)
  
  From what I can see, the Bb REST API is correctly adhering to REST standards, but I have used other APIs (Canvas') that extended REST a bit, or gave alternative ways to get data in larger batches. For example, something like /learn/api/public/v1/courses/{courseId}/users that would return user information for all users in a course in one REST call. Or possibly a way to embed User information inside each Course Membership (this does violate REST, though). With Canvas, we found that by leveraging these other alternative methods we improved performance considerably. With the {courseId}/users example, we could make simultaneous requests for Course Memberships and Users, then merge them, instead of waiting for the Course Memberships call to return to request Users.
  Like • Show 0 Likes0
  Actions

How to optimize REST API usage?

Outcomes

Related Content

Recommended Content

Incoming Links